Re: Euphoria's strengths
- Posted by jimcbrown (admin) Jan 12, 2015
- 1440 views
But adding the ability to dynamically add and remove variables from the variable table on the fly? In a translated program? It can be done, it's just a lot more work.
Sounds like we'd have to go to JIT for this to work.
Matt
Just to be clear, if the var is accessable for changing thru the normal existing syntax, the var table is then just for searching to gather var names, and see if they are init'd, etc, not to alter them or instanciate/remove new/old ones. At least that's my desire.
Ok, then that's much easier.
Today, debug.e already supports simple lookup of variable names via the symbol_lookup() function and friends. Wildcards and such aren't supported, however.
debug.e also already supports checking if a variable is init'd or not, via the is_novalue() function.
If this is instead of having string execution, i'd rather have the string execution first.
I don't see them being linked. This is a desired feature, but adding it is just plain hard. If someone handed us a patch with this already implemented (even interpreter only), we'd almost certainly take it and add it. However, the only person who has done this successfully that I know of is mattlewis, and his version is out of date (being based on a 2.x version of Euphoria) and hard to upgrade to the latest (due to it being based on a complete rewrite of the interpreter originally by dcuny).
i would be very happy to see code appear as interpreted-only, just to get it.
Then it's possible that what you want is already available today, in debug.e ...
I have not bound or compiled Eu in over a decade. Of course, i may want to again some day, and i understand why the interpreted and compiled (and bound?) should act the same. Ummm, am i correct that if a feature runs only interpreted, it runs when bound too?
No... binding does lose some information that an unbound program would have. Most features will work regardless, but getting all the metadata in a bound program is more work.
